From 239fa712c8f221c4b5fed797dcf506022fb69f89 Mon Sep 17 00:00:00 2001
From: Chris Mostert <15890652+chrismostert@users.noreply.github.com>
Date: Wed, 10 Apr 2024 11:57:03 +0200
Subject: [PATCH] Format reporting unit name to exclude zip code and prefix

---
 src/csv_write.py | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/src/csv_write.py b/src/csv_write.py
index 7df02a8..3717bad 100644
--- a/src/csv_write.py
+++ b/src/csv_write.py
@@ -1,4 +1,5 @@
 import csv
+import re
 from typing import Dict, List, Optional
 
 from eml import EML, CheckResult
@@ -21,6 +22,9 @@
 ]
 PROTOCOL_VERSION = "EP2024"
 
+ZIP_CODE_PATTERN = re.compile(r"\(postcode: \d{4} [A-Z]{2}\)")
+STEMBUREAU_PREFIX_PATTERN = re.compile(r"^Stembureau Stembureau")
+
 
 def _write_header(writer, metadata: EmlMetadata, description: str) -> None:
     writer.writerow(["Versie controleprotocol", PROTOCOL_VERSION])
@@ -68,6 +72,16 @@ def _format_percentage_deviation(percentage: float) -> str:
     return f"{sign}{percentage_int}%"
 
 
+def _format_reporting_unit_name(reporting_unit_name: Optional[str]) -> str:
+    return (
+        STEMBUREAU_PREFIX_PATTERN.sub(
+            "Stembureau", ZIP_CODE_PATTERN.sub("", reporting_unit_name)
+        ).strip()
+        if reporting_unit_name
+        else ""
+    )
+
+
 def _id_cols(metadata: EmlMetadata, id: str) -> List[Optional[str]]:
     return [
         metadata.election_id,
@@ -75,7 +89,7 @@ def _id_cols(metadata: EmlMetadata, id: str) -> List[Optional[str]]:
         metadata.authority_id,
         metadata.authority_name,
         _format_id(id),
-        metadata.reporting_unit_names.get(id),
+        _format_reporting_unit_name(metadata.reporting_unit_names.get(id)),
     ]